[aerohead]

The paper I have is from the 8th International Conference on Advanced Concepts in Mechanical Engineering, IOP Conference Series: Materials Science and Engineering 444, 2018.
--------------------------------------------------------------------------------------
And it's only for rear wheel arches.
--------------------------------------------------------------------------------------
The study was in CFD, using the Lattice Boltzman simplified microscopic mathematical method.... particle velocity distribution function, in discrete time steps, time-averaged motion, consistent with the Navier-Stokes equation.
--------------------------------------------------------------------------------------
The virtual model is a hatchback with 195/ 55R-16 tires, with Cd 0.377.
1) with the gap closed down to the outer diameter of the tire, the drag dropped to Cd 0.367, delta- 0.011 ( 2.65% ).
2) with the gap closed down to the diameter of the 'rim/wheel', the drag dropped to Cd 0.364, delta - 0.013, ( 3.44% ).
3) with a full skirt, delta- 0.024, ( 6.36% )
-------------------------------------------------------------------------------------
In 2012,Coventry University studied an Audi A3 hatchback, Cd 0.316.
1) rear skirts gave Cd 0.310, delta- 0.006, ( 1.89% ).
2) front skirts gave Cd 0.303, delta- 0.013, ( 4.1% ).
3) front & rear skirts gave Cd 0.294, delta- 0.022 ( 6.96% )
--------------------------------------------------------------------------------------
The Audi, a 1936 Adler-Jaray, and 1983 Ford Probe-IV all indicated for a greater drag reduction for a modified front wheel, compared to their respective rear wheels.
Front gap-fillers might trend the same way.
Additional research would help nail that down.